Tissue-defining transcriptional signatures

In this analysis I explore transcriptional signatures that are associated with each of the tissue sites. The idea was to perform a rather general differential expression analysis in healthy individuals and then perform some sort of clustering analysis to define tissue-defining gene signatures. The aim is really to see if we can see what functional differences there are between intestinal sites and how these may be affected in inflammatory bowel disease.

Analysis overview

In previous analyses we have observed that tissue expression profiles are very different. Particularly in PCA, the Ileum samples cluster away from caecum and rectum on PC1. Rectum and Caecum can be distinguished based on PC3. Therefore, here I perform DESeq2 differential abundance analysis between tissues. This is done in heathy individuals so that there are no influences of disease state on tissue profiles.

The data set at a glance

This is a summary of the data set that I am working with. I will base some of the filtering steps on the number of samples that I have in each tissue/disease group. Below is a summary of these numbers.

	HEALTHY	PSC/UC	UC
Caecum	9	7	10
Ileum	11	7	9
Rectum	12	7	10

The lowest number of samples for any group is 7 PSC/UC individuals in the Rectum. When doing differential expression analysis across tissues I set a threshold of keeping genes that have at least 10 reads in at least 7 individuals. This is so that the definition of tissue-defining signatures can be compared directly when I go on to perform differential expression analysis between diseases and healthy controls.

Principle components analysis on healthy samples

I have already performed this analysis during exploration of the data. Here as I am performing tissue differential expression on samples from healthy individuals I remove PSC/UC and UC samples before running the PCA.

Differential expression results

I run the LRT function in DESeq2 in order to define genes that are differentially expressed between any of the tissues. The idea is that genes that show similar patterns of expression across tissues will then be defined into signatures via cluster analysis.

Dispersion estimates

Below is the plot of the dispersion estimates obtained from DESeq2. DESeq2 is run controlling for Patient.ID using the LRT test. The resulting differentially expressed set is therefoer any gene that varies in any way across tissue sites.

Differentially expressed genes

After runnng DESeq2 I found there were 13070 significant differences found between the tissues.

Visualising differences

Here I visualise differences across tissues in a heatmap.

There are a large number of significantly differentially expressed genes across tissue sites. They appear to fall into fairly neat clusters which I will expore in teh next sections.

kmeans analysis

The above shows us how genes vary between any of the tissues. As can be seen from the heatmap there is some very nice clustering by tissue. Although we can visually identify distinct clusters of genes that are differentially expressed between tissues, first we use the elbow method to determine the optimal number of clusters.

Elbow plot

The issue with the above plot is that it is quite a smooth curve and there isn’t such a definitive cut-off. I therefore will explore the use of the dynamic tree cut that will use the same parameters for clustering as in the heatmap and may give a suitable defintion for gene clusters.

Dynamic tree cut analysis

Here I use the dynamic tree cutting method to place genes into clusters.

##  ..done.

The dynamic tree cutting method identified 5. These are annotated on the heatmap.

## png 
##   3

## png 
##   3

These 5 clusters look to represent the data pretty well - at least by eye. I will use these clusters to perform pathways analysis.

GO biological Pathways analysis by cluster

Here I run pathways analysis (GO biological pathways) on each cluster to determine which pathways are enriched in the different tissues. Top 10 pathways in each cluster are plotted (i.e. top 10 significantly enriched pathways). This is performed using cgat-apps runGO.py.

Cluster 1: High in Ileum

Top 10 enriched pathways in cluster 1

	code	scount	stotal	spercent	bcount	btotal	bpercent	ratio	pvalue	pover	punder	goid	category	description	fdr
27	+	64	2674	2.39	138	12394	1.11	2.15	0	0	1	GO_CELLULAR_LIPID_CATABOLIC_PROCESS	GO_BP	GO_CELLULAR_LIPID_CATABOLIC_PROCESS	0
28	+	251	2674	9.39	793	12394	6.40	1.47	0	0	1	GO_CELLULAR_LIPID_METABOLIC_PROCESS	GO_BP	GO_CELLULAR_LIPID_METABOLIC_PROCESS	0
87	+	89	2674	3.33	208	12394	1.68	1.98	0	0	1	GO_LIPID_CATABOLIC_PROCESS	GO_BP	GO_LIPID_CATABOLIC_PROCESS	0
91	+	312	2674	11.67	985	12394	7.95	1.47	0	0	1	GO_LIPID_METABOLIC_PROCESS	GO_BP	GO_LIPID_METABOLIC_PROCESS	0
111	+	154	2674	5.76	429	12394	3.46	1.66	0	0	1	GO_MONOCARBOXYLIC_ACID_METABOLIC_PROCESS	GO_BP	GO_MONOCARBOXYLIC_ACID_METABOLIC_PROCESS	0
140	+	252	2674	9.42	827	12394	6.67	1.41	0	0	1	GO_ORGANIC_ACID_METABOLIC_PROCESS	GO_BP	GO_ORGANIC_ACID_METABOLIC_PROCESS	0
144	+	147	2674	5.50	417	12394	3.36	1.63	0	0	1	GO_ORGANIC_HYDROXY_COMPOUND_METABOLIC_PROCESS	GO_BP	GO_ORGANIC_HYDROXY_COMPOUND_METABOLIC_PROCESS	0
215	+	257	2674	9.61	838	12394	6.76	1.42	0	0	1	GO_SINGLE_ORGANISM_CATABOLIC_PROCESS	GO_BP	GO_SINGLE_ORGANISM_CATABOLIC_PROCESS	0
219	+	107	2674	4.00	287	12394	2.32	1.73	0	0	1	GO_SMALL_MOLECULE_CATABOLIC_PROCESS	GO_BP	GO_SMALL_MOLECULE_CATABOLIC_PROCESS	0
220	+	426	2674	15.93	1534	12394	12.38	1.29	0	0	1	GO_SMALL_MOLECULE_METABOLIC_PROCESS	GO_BP	GO_SMALL_MOLECULE_METABOLIC_PROCESS	0

Cluster 2: Low in Ileum

Top 10 enriched pathways in cluster 2

	code	scount	stotal	spercent	bcount	btotal	bpercent	ratio	pvalue	pover	punder	goid	category	description	fdr
11	+	242	2319	10.44	471	12394	3.80	2.75	0	0	1	GO_AMIDE_BIOSYNTHETIC_PROCESS	GO_BP	GO_AMIDE_BIOSYNTHETIC_PROCESS	0
12	+	30	2319	1.29	51	12394	0.41	3.14	0	0	1	GO_AMINO_ACID_ACTIVATION	GO_BP	GO_AMINO_ACID_ACTIVATION	0
37	+	244	2319	10.52	933	12394	7.53	1.40	0	0	1	GO_CARBOHYDRATE_DERIVATIVE_METABOLIC_PROCESS	GO_BP	GO_CARBOHYDRATE_DERIVATIVE_METABOLIC_PROCESS	0
40	+	277	2319	11.94	654	12394	5.28	2.26	0	0	1	GO_CELLULAR_AMIDE_METABOLIC_PROCESS	GO_BP	GO_CELLULAR_AMIDE_METABOLIC_PROCESS	0
45	+	153	2319	6.60	465	12394	3.75	1.76	0	0	1	GO_CELLULAR_COMPONENT_DISASSEMBLY	GO_BP	GO_CELLULAR_COMPONENT_DISASSEMBLY	0
49	+	209	2319	9.01	623	12394	5.03	1.79	0	0	1	GO_CELLULAR_MACROMOLECULAR_COMPLEX_ASSEMBLY	GO_BP	GO_CELLULAR_MACROMOLECULAR_COMPLEX_ASSEMBLY	0
50	+	299	2319	12.89	1142	12394	9.21	1.40	0	0	1	GO_CELLULAR_MACROMOLECULE_LOCALIZATION	GO_BP	GO_CELLULAR_MACROMOLECULE_LOCALIZATION	0
53	+	74	2319	3.19	118	12394	0.95	3.35	0	0	1	GO_CELLULAR_PROTEIN_COMPLEX_DISASSEMBLY	GO_BP	GO_CELLULAR_PROTEIN_COMPLEX_DISASSEMBLY	0
54	+	70	2319	3.02	135	12394	1.09	2.77	0	0	1	GO_CELLULAR_RESPIRATION	GO_BP	GO_CELLULAR_RESPIRATION	0
68	+	323	2319	13.93	1170	12394	9.44	1.48	0	0	1	GO_CELL_CYCLE	GO_BP	GO_CELL_CYCLE	0

Cluster 3: High in Rectum

Top 10 enriched pathways in cluster 3

	code	scount	stotal	spercent	bcount	btotal	bpercent	ratio	pvalue	pover	punder	goid	category	description	fdr
140	+	255	1728	14.76	1191	12394	9.61	1.54	0e+00	0e+00	1	GO_NEUROGENESIS	GO_BP	GO_NEUROGENESIS	0.0e+00
142	+	139	1728	8.04	585	12394	4.72	1.70	0e+00	0e+00	1	GO_NEURON_DEVELOPMENT	GO_BP	GO_NEURON_DEVELOPMENT	0.0e+00
143	+	168	1728	9.72	728	12394	5.87	1.66	0e+00	0e+00	1	GO_NEURON_DIFFERENTIATION	GO_BP	GO_NEURON_DIFFERENTIATION	0.0e+00
50	+	83	1728	4.80	315	12394	2.54	1.89	0e+00	0e+00	1	GO_CELL_MORPHOGENESIS_INVOLVED_IN_NEURON_DIFFERENTIATION	GO_BP	GO_CELL_MORPHOGENESIS_INVOLVED_IN_NEURON_DIFFERENTIATION	2.0e-06
43	+	133	1728	7.70	594	12394	4.79	1.61	0e+00	0e+00	1	GO_CELL_CELL_SIGNALING	GO_BP	GO_CELL_CELL_SIGNALING	3.0e-06
48	+	230	1728	13.31	1176	12394	9.49	1.40	0e+00	0e+00	1	GO_CELL_DEVELOPMENT	GO_BP	GO_CELL_DEVELOPMENT	5.0e-06
149	+	86	1728	4.98	345	12394	2.78	1.79	0e+00	0e+00	1	GO_NEURON_PROJECTION_MORPHOGENESIS	GO_BP	GO_NEURON_PROJECTION_MORPHOGENESIS	9.0e-06
232	+	48	1728	2.78	160	12394	1.29	2.15	1e-07	1e-07	1	GO_REGULATION_OF_NEUROTRANSMITTER_LEVELS	GO_BP	GO_REGULATION_OF_NEUROTRANSMITTER_LEVELS	2.7e-05
277	+	80	1728	4.63	326	12394	2.63	1.76	2e-07	2e-07	1	GO_SYNAPTIC_SIGNALING	GO_BP	GO_SYNAPTIC_SIGNALING	3.8e-05
148	+	50	1728	2.89	174	12394	1.40	2.06	2e-07	2e-07	1	GO_NEURON_PROJECTION_GUIDANCE	GO_BP	GO_NEURON_PROJECTION_GUIDANCE	5.5e-05

Cluster 4: Low in Rectum

Top 10 enriched pathways in cluster 4

	code	scount	stotal	spercent	bcount	btotal	bpercent	ratio	pvalue	pover	punder	goid	category	description	fdr
2	+	101	1160	8.71	394	12394	3.18	2.74	0	0	1	GO_ACTIVATION_OF_IMMUNE_RESPONSE	GO_BP	GO_ACTIVATION_OF_IMMUNE_RESPONSE	0
3	+	83	1160	7.16	244	12394	1.97	3.63	0	0	1	GO_ADAPTIVE_IMMUNE_RESPONSE	GO_BP	GO_ADAPTIVE_IMMUNE_RESPONSE	0
4	+	49	1160	4.22	141	12394	1.14	3.71	0	0	1	GO_ADAPTIVE_IMMUNE_RESPONSE_BASED_ON_SOMATIC_RECOMBINATION_OF_IMMUNE_RECEPTORS_BUILT_FROM_IMMUNOGLOBULIN_SUPERFAMILY_DOMAINS	GO_BP	GO_ADAPTIVE_IMMUNE_RESPONSE_BASED_ON_SOMATIC_RECOMBINATION_OF_IMMUNE_RECEPTORS_BUILT_FROM_IMMUNOGLOBULIN_SUPERFAMILY_DOMAINS	0
10	+	48	1160	4.14	184	12394	1.48	2.79	0	0	1	GO_ANTIGEN_RECEPTOR_MEDIATED_SIGNALING_PATHWAY	GO_BP	GO_ANTIGEN_RECEPTOR_MEDIATED_SIGNALING_PATHWAY	0
12	+	174	1160	15.00	899	12394	7.25	2.07	0	0	1	GO_BIOLOGICAL_ADHESION	GO_BP	GO_BIOLOGICAL_ADHESION	0
20	+	33	1160	2.84	87	12394	0.70	4.05	0	0	1	GO_B_CELL_MEDIATED_IMMUNITY	GO_BP	GO_B_CELL_MEDIATED_IMMUNITY	0
22	+	24	1160	2.07	50	12394	0.40	5.13	0	0	1	GO_B_CELL_RECEPTOR_SIGNALING_PATHWAY	GO_BP	GO_B_CELL_RECEPTOR_SIGNALING_PATHWAY	0
41	+	111	1160	9.57	503	12394	4.06	2.36	0	0	1	GO_CELL_ACTIVATION	GO_BP	GO_CELL_ACTIVATION	0
42	+	33	1160	2.84	115	12394	0.93	3.07	0	0	1	GO_CELL_ACTIVATION_INVOLVED_IN_IMMUNE_RESPONSE	GO_BP	GO_CELL_ACTIVATION_INVOLVED_IN_IMMUNE_RESPONSE	0
43	+	102	1160	8.79	522	12394	4.21	2.09	0	0	1	GO_CELL_CELL_ADHESION	GO_BP	GO_CELL_CELL_ADHESION	0

Cluster 5: Subset of ileum samples high in expression

No significant pathways for this small cluster.

Matrices of genes in top ten pathways for each cluster

This set of analyses serves to look at how pathways related to each other in terms of the genes that are present (and differentially expressed). I provides a guage of how redundant the significant biological pathway gene sets are.

Cluster 1

Cluster 2

Cluster 3

Cluster 4

Genes of interest based on pathways analysis

Is is of interest to drill down a little deeper into genes that contribute to the significantly enriched pathways. As lipid metabolism and various other metabolism-based pathways came out I will use the reactome database annotations to try and find some more meaningful (i.e. lower level functions) and lower level annotations for genes of interest.

Reactome analysis

I perfomr enrchment analysis of reactome genesets here.

Examining the results tables suggests that there are a number of solute carriers that are differentially expressed between tissues. I next have a look at whether there is any enrichment of any particular family of SLC proteins. This requires building a database with the family memebers annotated and then running the enrichment analysis.

There isn’t much going on in terms of specific families of SLC transporters - especially in the Ileum. However there are the Ileal bile acid transporters that are in cluster 1 (ileum). Bile acid uptake is important in regulating nutrient absorbtion in the small intestine. It is of interest that bile acids activate FXR and we do see that both FXR1 and FXR2 are part of cluster1, suggesting that these have a role in the ileum. These nuclear receptors are likely acting to regulate lipid metabolism in the ileum. Below I plot the bile acid transporters and FXR1 and FXR2 to show that bile acid related pathways are different in the ileum compared with the caecum and rectum.

Bile acid transport and function in the ileum

Below are the plots showing the bile acid transporters and FXR nuclear receptors across tissues.

Cytochrome P450 components

A number of cytochrome P450 components fall into cluster 1. These are part of small molecule metabolic pathways and are involved in xenobiotic metabolism. It is unsurprising that these are more highly expressed in the ileum compared to the large intestine. Visualisation of these components is below.

ATP synthase components

Of note, cluster 2 contains mitochondrial genes involved in OXPHOS - this includes a number of ATP synthase genes plotted below.

Protein translation

Also of note, cluster 2 contains genes involved in translation e.g. ribosomal proteins involved in the GO_AMIDE_BIOSYNTHETIC_PROCESS. This suggests cellular growth (I think). Of interest and based on other observations, there are a number of immune-cell specific ribosomal proteins that were defined in this paper

SNARE complex and neuronal processes

Cluster 3, whish is predominantly rectum-associated, has higher expression of genes involved in synapse assembly and formation. It is likely that this represents an increased frequency of nneuronal cell types in the rectum relative to other cell types.

Immune pathways

Cluster 4 represents activated immune signatures. The cluster itself is a bit of a mix between genes that are more highly expressed in the caecum/rectum and those that are higher in the ileum. Clearly there is a B-cell and adaptive immune system signature. Below I plot the B cell signature (GO_B_CELL_MEDIATED_IMMUNITY).

Enrichment of transcription factor motifs in clusters

Given the large variation in gene expression between tissue sites it was of interest to see if this translates to differences in transcritpion factor signatures that may be responsible for tissue-defining gene clusters. To determine this, I run enrichment analysis using transfac match genesets from the MsigDB (v6.1).

Below are plots to show the top ten motifs that are enriched in the different gene clusters.

Cluster 1 tf enrichment

Cluster 2 tf enrichment

Cluster 3 tf enrichment

Cluster 4 tf enrichment

Genes of interest

It is difficult to narrow down exactly which genes to present and which are the most important in each cluster. Here I look at how genes that fall into the different significant GO biological pathways are related to genes that are part of enriched transcription factor motifs. That is, the expectation is that enriched transcription factor motifs will be involved in some of the biological pathways that are also associated.

HNF1 and lipid metabolism

There are established links between HNF transcription factors and the regulation of fatty acid metabolism. Here I look at a selection of genes that link HNF and lipid metabolism. For example see here

Protein synthesis, proliferation etc

E2F transcription factor is an elongation factor that is important for regulation of cell cycle genes. It is unsurprising that this transcription factor has emerged as regulating cluster 2 genes. Here we investigate how it relates to cell cycle control.

SRF transcription factor and neuronal functions?

SRF transcription factors look to be involved in neuronal survival although it has broader functions in multiple cells types

Neurotransmitter release in the rectum

From the data above it seems like there is increased neurogenesis and/or neuronal survival. Although not shown here, there is no general increased expression of the neuronal marker ELAVL4, suggesting that there isn’t an increased density of neurons in the rectum. It may be that there is an increase in a specific sub-population of neurons and it therefore may be helpful to look at the set of genes that are involved in neurotransmitter release to get an idea of this.

ETS transcription factor and adaptive immunity

ETS transcription factors are broadly expressed and involved in multiple cellular processes. It has also been shown that ETS transcription factors are also involved in activation and proliferative capacity of T cells. It is also expressed at high levels in B cells. Here I hypothesise that the ETS is regulating genes involved in lymphocyte activation in cluster 4 (high in caecum predominantly).

Immunoglobulin genes

It is quite difficult to work out what is going on with the Ig genes but I want to include them here to think about.

Factors associated with tertiary lymphoid structure formation

Emily suggested to look at the single cell paper from Alison Simmons group that describes a population of cells that are especially present (S4) in UC patient biopsies that look to be involved in recruiting lymphocytes and follicle formation. I am interested to see which sites are more likely to have follicles based on these markers.

It’s still a bit of a mixed bag here although activated stroma e.g PDPN is more highly expressed in the caecum and falls in the “immune” cluster.

Correlation betwen Ig gene expression and proliferation markers

I am interested in whether the increased Ig expression seen in the caecum is somehow related to proliferation of B-cells. Here I look at the correlation between Ig gene expression and the cell cycle control gene E2F1.

The only significant correlation was a negative one with the IGBP1. This may be of interest as this protein binds IgM. It is more highly expressed in the Ileum - maybe as B cells have not class switched here??

I produced the plot above to see if there was a correlation between IGBP1 and cell cycle genes i.e. to see if its associated with proliferation. It is not when you account for the different tissues.

Correlation with lymphocyte-specific cell-cycle control genes

E2F1 is quite a general cell-cycle control gene. I suspect that we are looking at increased proliferation in the caecum of a subset of cells - most likely B-cells. In order to see whether this is likely to be the case, I look at the correlation between Ig genes and the lymphocyte-specific cell-cylce control gene HELLS (lymphocyte-specific helicase).

	cor	p.value	padj
ENSG00000089289	-0.6116283	0.0001998	0.0331618

Again, the only significant correlation was a negative one with IGBP1. It’s quite difficult to draw any conclusions from this relationship as I can’t find anything much in the literature about the specificity of IGBP1.

Writing genesets

In further analysis I will explore how differentially expressed genes between disease and healthy fall in terms of the clusters defined above. Therfore here I produce genesets for input into that analysis. These are written out here.

##  [1] "R version 3.6.1 (2019-07-05)"                                                                        
##  [2] "Platform: x86_64-conda_cos6-linux-gnu (64-bit)"                                                      
##  [3] "Running under: CentOS release 6.7 (Final)"                                                           
##  [4] ""                                                                                                    
##  [5] "Matrix products: default"                                                                            
##  [6] "BLAS/LAPACK: /gfs/devel/nilott/cgat-24102019/conda-install/envs/cgat-flow/lib/libopenblasp-r0.3.9.so"
##  [7] ""                                                                                                    
##  [8] "locale:"                                                                                             
##  [9] " [1] LC_CTYPE=en_GB.UTF-8       LC_NUMERIC=C              "                                          
## [10] " [3] LC_TIME=en_GB.UTF-8        LC_COLLATE=en_GB.UTF-8    "                                          
## [11] " [5] LC_MONETARY=en_GB.UTF-8    LC_MESSAGES=en_GB.UTF-8   "                                          
## [12] " [7] LC_PAPER=en_GB.UTF-8       LC_NAME=C                 "                                          
## [13] " [9] LC_ADDRESS=C               LC_TELEPHONE=C            "                                          
## [14] "[11] LC_MEASUREMENT=en_GB.UTF-8 LC_IDENTIFICATION=C       "                                          
## [15] ""                                                                                                    
## [16] "attached base packages:"                                                                             
## [17] " [1] grid      parallel  stats4    stats     graphics  grDevices utils    "                          
## [18] " [8] datasets  methods   base     "                                                                  
## [19] ""                                                                                                    
## [20] "other attached packages:"                                                                            
## [21] " [1] ggrepel_0.8.2               cluster_2.1.0              "                                        
## [22] " [3] dplyr_0.8.0.1               reshape_0.8.8              "                                        
## [23] " [5] data.table_1.12.8           gtools_3.8.2               "                                        
## [24] " [7] dynamicTreeCut_1.63-1       vegan_2.5-6                "                                        
## [25] " [9] lattice_0.20-41             permute_0.9-5              "                                        
## [26] "[11] ggplot2_3.2.1               pheatmap_1.0.12            "                                        
## [27] "[13] RSQLite_2.1.2               vsn_3.54.0                 "                                        
## [28] "[15] DESeq2_1.26.0               SummarizedExperiment_1.16.0"                                        
## [29] "[17] DelayedArray_0.12.0         BiocParallel_1.20.0        "                                        
## [30] "[19] matrixStats_0.56.0          Biobase_2.46.0             "                                        
## [31] "[21] GenomicRanges_1.38.0        GenomeInfoDb_1.22.0        "                                        
## [32] "[23] IRanges_2.20.0              S4Vectors_0.24.0           "                                        
## [33] "[25] BiocGenerics_0.32.0         knitr_1.25                 "                                        
## [34] "[27] gridExtra_2.3               gplots_3.0.3               "                                        
## [35] "[29] RColorBrewer_1.1-2         "                                                                    
## [36] ""                                                                                                    
## [37] "loaded via a namespace (and not attached):"                                                          
## [38] " [1] nlme_3.1-147           bitops_1.0-6           bit64_0.9-7           "                           
## [39] " [4] tools_3.6.1            backports_1.1.6        affyio_1.56.0         "                           
## [40] " [7] R6_2.4.1               rpart_4.1-15           KernSmooth_2.23-16    "                           
## [41] "[10] mgcv_1.8-31            Hmisc_4.2-0            DBI_1.0.0             "                           
## [42] "[13] lazyeval_0.2.2         colorspace_1.4-1       nnet_7.3-13           "                           
## [43] "[16] withr_2.1.2            tidyselect_1.0.0       preprocessCore_1.48.0 "                           
## [44] "[19] bit_1.1-15.2           compiler_3.6.1         htmlTable_1.13.3      "                           
## [45] "[22] labeling_0.3           caTools_1.18.0         scales_1.1.0          "                           
## [46] "[25] checkmate_2.0.0        affy_1.64.0            genefilter_1.68.0     "                           
## [47] "[28] stringr_1.4.0          digest_0.6.25          foreign_0.8-76        "                           
## [48] "[31] rmarkdown_1.16         XVector_0.26.0         base64enc_0.1-3       "                           
## [49] "[34] jpeg_0.1-8.1           pkgconfig_2.0.3        htmltools_0.4.0       "                           
## [50] "[37] highr_0.8              limma_3.42.0           htmlwidgets_1.5.1     "                           
## [51] "[40] rlang_0.4.5            rstudioapi_0.11        farver_2.0.3          "                           
## [52] "[43] acepack_1.4.1          RCurl_1.95-4.12        magrittr_1.5          "                           
## [53] "[46] GenomeInfoDbData_1.2.1 Formula_1.2-3          Matrix_1.2-18         "                           
## [54] "[49] Rcpp_1.0.4.6           munsell_0.5.0          lifecycle_0.2.0       "                           
## [55] "[52] stringi_1.4.3          yaml_2.2.1             MASS_7.3-51.4         "                           
## [56] "[55] zlibbioc_1.32.0        plyr_1.8.6             blob_1.2.1            "                           
## [57] "[58] gdata_2.18.0           crayon_1.3.4           splines_3.6.1         "                           
## [58] "[61] annotate_1.64.0        locfit_1.5-9.4         pillar_1.4.3          "                           
## [59] "[64] geneplotter_1.64.0     XML_3.98-1.20          glue_1.4.0            "                           
## [60] "[67] evaluate_0.14          latticeExtra_0.6-29    BiocManager_1.30.10   "                           
## [61] "[70] png_0.1-7              vctrs_0.2.4            gtable_0.3.0          "                           
## [62] "[73] purrr_0.3.4            assertthat_0.2.1       xfun_0.13             "                           
## [63] "[76] xtable_1.8-4           survival_3.1-12        tibble_3.0.1          "                           
## [64] "[79] AnnotationDbi_1.48.0   memoise_1.1.0          ellipsis_0.3.0        "